Reconstruction of biological networks is a crucial step in extracting information from a large volume of experimental data. Various methods have been developed to reconstruct networks from data, each of which possesses its own strength and disadvantages. In this talk, I will first describe examples of network reconstruction techniques including classic optimization-based approaches (e.g., least-squares methods), dimensionality reduction methods (e.g., statistical significance tests combined with either principal components regressions (PCR), or partial least squares (PLS), Bayesian networks and hybrid methods (e.g., Linear Matrix Inequalities (LMI) and Least Absolute Shrinkage and Selection Operator (LASSO). Next, I will introduce our most recent method, called Doubly Penalized Linear Absolute Shrinkage and Selection Operator (DPLASSO) for network reconstruction with the intent to combine the beneficial features of a regression and statistical significance testing-based method and a penalized optimization method. I will present results from applications to simulated data from synthetic random networks as well as from a biological system, namely the cell cycle in budding yeast.

Background papers:

1. Behrang Asadi, Mano R. Maurya, Daniel Tartakovsky, Shankar Subramaniam, Comparison of Statistical andOptimisation-based Methods for Data-driven Network Reconstruction of Biological Systems , IET Systems Biology,2012,6(5).
   \t
2. Robert J.Prill, JulioSaez-Rodriguez, Leonidas G.Alexopoulos, Peter K.Sorger, GustavoStolovitzky, Crowdsourcing Network Inference: The DREAM Predictive Signaling Network Challenge . Science Signaling, 2011.4(189).
   \t
3. Behrang Asadi, Mano R. Maurya, Daniel Tartakovsky, Shankar Subramaniam, Doubly Penalized LASSO for Reconstruction of Biological Networks, 2012 IEEE Second Conference on Healthcare Informatics, Imaging and Systems Biology, San Diego, p. 129, 27-28 Sept. 2012.